Modeling Key Parameters for Greenhouse using Fuzzy Clustering Technique

Efren Gorrostieta Hurtado, Artemio Sotomayor

Olmedo, Jesus Carlos Pedraza Ortega.
CIDIT-Facultad de Informtica, Campus Juriquilla
Universidad Autnoma de Quertaro
Quertaro, Mxico.
[email protected],
[email protected], [email protected].

Marco Antonio Aceves Fernandez, Ubaldo Geovanni
Villaseor Carillo
Facultad de Informtica, Campus Juriquilla
Universidad Autnoma de Quertaro
Quertaro, Mxico
[email protected], [email protected].
Abstract The clustering techniques are usually used in
classification and pattern recognition. Moreover, fuzzy logic is
used for system modeling when the information is scarce,
inaccurate or its behavior is described using a complex
mathematical model. As example of this type of system, a
greenhouse is considered, where the variables are: in-house
and out-house temperature, humidity for both inside and
outside the greenhouse and wind direction. These variables
show a dynamic and non-linear behavior; being the in-house
temperature and internal humidity the variables of concern for
the greenhouse control and modeling. In this project, the
development and implementation of three clustering
algorithms, being fuzzy K-means, Fuzzy C-means and fuzzy
clustering subtractive, is presented. This project is used as the
foundation for the design of fuzzy systems and its application
in temperature and humidity modeling of a greenhouse used as
a laboratory of biotronics at the Universidad Autonoma de
Queretaro.
Keywords; Clustering; Pattern Recognition; Fuzzy logic;
modeling.
I. INTRODUCTION
During the last two decades, a large effort was devoted to
develop adequate greenhouse climate and crop models, for
simulation, control and management purposes [1-5]. The
study and design of efficient greenhouse environmental
controllers require having a priori knowledge of the
greenhouse climate models. These models must be related
with the external influences of the outside weather conditions
(such as solar radiation, outside air temperature, wind
velocity, etc.), and with the actuating actions performed
(such as ventilation, cooling, heating, among others). The
practical goal of this work is to model the greenhouse air
temperature and humidity using clustering techniques and
made an automatically generator of fuzzy rules relations
from real data in order to predict the behavior inside the
greenhouse.
II. FUZZY SYSTEMS
Since 1985 when the fuzzy model methodology
suggested by Takagi-Sugeno [5][6], as well known as the
(Takagi-Sugeno-Khan) TSK model, has been widely applied
on theoretical analysis, control applications and fuzzy
modeling.
Fuzzy system needs the antecedent and consequence to
express the logical connection between the input-data and
output-data that is used as a basis to produce the desired
system behavior [6]. Fuzzy systems provide a simple,
flexible way to express knowledge by using fuzzy IF- THEN
Rules [5] and fuzzy modeling provide a framework for
modeling complex non-linear relations, compared to
traditional mathematical modeling, fuzzy modeling
possesses some distinctive advantages, such as the
mechanism of reasoning in human understandable terms, the
capacity of taking linguistic information from human experts
and combining it with numerical data and the ability of
approximating complex non-linear functions with simple
models[5-10]. A set of n fuzzy rules performs a fuzzy
system. The IF-THEN are the basis of TSK and any other
kind of fuzzy system and are given by:

R
i
= IF Y
1
is A
1
and Y
2
is A
2
and
THEN Z
1
is B
1
and Z
2
is B
2

(1)

where Ri is ith fuzzy rule and i pertenecent to set of n fuzzy
rules. In order to automatically generate a set of fuzzy rules,
clustering techniques are applied where, each cluster center
can be used as the basis of a rule that describes the system
behavior. In this survey, Fuzzy K-means, Fuzzy C-means
Clustering and Fuzzy clustering subtractive are used.

III. CLUSTERING TECHNIQUES

Clustering techniques have been widely used to generate
groups of subsets of data in which a group called cluster is
dense in the sense that a distance within a group is small,
whereas a distance between clusters is sparse in that two
objects from different clusters are distant [8][11]. Another
fundamental idea is to use fuzzy sets in clustering. The
fuzzy approach to clustering is capable of producing many
methods and algorithms although fuzzy system for the
Ninth Mexican International Conference on Artificial Intelligence
978-0-7695-4284-3/10 $26.00 2010 IEEE
DOI 10.1109/MICAI.2010.37
103
present purpose does not have profound mathematical
structure in particular. The main reason that the fuzzy
approach has such capability is its inherent feature of
linking-connecting different methodologies including
statistical models, machine learning, and various other
heuristics [11-12].
A. K-means clustering

The best-known and earliest partition clustering technique,
the K-means clustering algorithm. In spite of having many
successful applications in several fields such pattern
recognition and system identification. The main characteristic
is that membership values being only 0 or 1 may not always
reflect the practical relationship between the data point and
the cluster; its algorithm is explained below.

Step 1: Initializae the cluster center c
i
, i=1, , c.
This is typically achived by random by a selection of c
ponts from hmong all of the data points.

Step 2: Determine the membership matrix U by
Equation:


(2)

Where, u
ij
are the cluster center elements of the
matrix U, an the x
j
, son los are datapoints that belog to
each cluster center i, otherwise becomes 0.

Step 3: Calculate the cost function acording to
equation (3). Stop if either it is below certain tolerante
value or its improvement over previous iteration is
below a certain threshold :


(3)

The algorithm is iterative iterative, and no guarantee can
be made that it will converge to an optimum solution, in
order to overcome this drawback the k-means algorithm can
be modified, this modifications is generally called on-line
mode, where the cluster centers an the corresponding groups
are derived through time averaging. That is, for a given data
point x, the algorithm finds the closest cluster center c
i
and i s
updated using the formula.

!c
i
= "(x # c
i
) (4)

This formula is essentially embedded in many learning
rules on unsupervised learning via neural Networks [9].

B. Fuzzy c-means clustering

Fuzzy C-Means clustering (FCM) is an iterative
optimization algorithm that minimizes the cost function
given by:



Where n is the number of data points, c is the number of
clusters, x
k
is the kth data point, v
i
is the ith cluster center
ik
is
the degree of membership of the kth data in the ith cluster, and
m is a constant greater than 1 (typically m=2)[12]. The degree
of membership
ik
is defined by:

!
=
"
"
"
=
c
j
m
i k
i k
ik
v x
v x
1
) 1 /( 2
)
|| ||
|| ||
(
1

(6)

Starting with a desired number of clusters c and an initial
guess for each cluster center v
i
, i = 1,2,3 c, FCM will
converge to a solution for v
i
that represents either a local
minimum or a saddle point cost function [10]. The FCM
method employs fuzzy partitioning such that each point can
belong to several clusters with membership values between 0
and 1. FCM include predefined parameters such as the
weighting exponent m and the number of clusters c.
C. Fuzzy Subtractive Clustering
The subtractive clustering method assumes each data
point is a potential cluster center and calculates a measure of
the likelihood that each data point would define the cluster
center, based on the density of surrounding data points.
Consider m dimensions n data point (x1,x2, ..., xn) and each
data point is potential cluster center, the density function Di
of data point at xi is given by:


(7)

where r
a
is a positive number. More data points surround the
data point with the highest potential. A radius defines a
neighbor area then the data points, which exceed ra, have no
influence on the density of data point.
After calculating the density function of each data point
is possible to select the data point with the highest potential
and find the first cluster center. Assuming that xc1 is selected
and Dc1 is its density, the density of each data point can be
amended by:


J =
k=1
n
!

ik
m
|| x
k
"v
i
||
2
i=1
c
!


(5)
104

(8)

The density function of data point, which is close to the first
cluster center, is reduced, so these data points cannot
become the next cluster center. rb defines an neighbor area
where the density function of data point is reduced. Usually
constant rb > ra. In order to avoid the overlapping of cluster
centers near to other(s) is given by [9-11]:


(9)

Once amending the density function of data points is
possible to find the next cluster center. Then repeat this
process until find all cluster centers. The subtractive
clustering is used to determine the number of clusters of the
data being proposed, and then generates a fuzzy model.

IV. FUZZY MODEL CONSTRUCTION

When cluster estimation method is applied to a
collection of input/output data, each cluster center
exemplifies a characteristic behavior of the system. Hence,
each cluster center can be used as the basis of a rule that
describes the system behavior. Consider a set of c cluster
centers

{x
1
*
, x
2
*
, x
3
*
,...x
c
*
} in an M-dimensional space. Let
the first N dimensions correspond to the input variables and
the last MN dimensional corresponds to output variables.
Each vector

x
i
*
could be decomposed into two component
vectors

y
i
*
and

z
i
*
where

y
i
*
contains the first N elements
of

x
i
*
and

z
i
*
contains the last MN elements
Then consider each cluster center

x
i
*
is represents a
fuzzy rule that describes the system behavior. Given an
input vector y, the degree to which rule i is fulfilled is
defined by:

i
= e
!"|| yy
i
*
||
2

(10)

Where ! is the constant defined by[9].


! =
4
r
a
2

(11)

and r
a
is a positive constant[9].

The output vector z is computed by

z =

i
z
i
*
i!1
c
"

i
i!1
c
"
(12)

Then this computational model is in terms of a fuzzy inference
system employing if-then rules following the Equation 1 form.
Where Y
1
is the jth input variable and Z
1
is the jth output
variable. A
1
is an exponential membership function and B
1
is a
singleton for the ith rule that is represented by cluster center

x
i
*
, A
j
and B
j
are given by:

A
j
(q) = e
!"(q!y
ij
*
)
2
(13)

B
j
= z
ij
*
(14)

Where

y
ij
*
is the jth element of

y
i
*
and

z
ij
*
is the jth element
of

z
i
*
. This computational scheme is equivalent to an inference
method that uses multiplication as the AND operator, weights
the output of each rule by the firing strength, and computes the
output as a weighted average of the output of each rule [8-13].

V. STUDY CASE AND EXPERIMENTAL RESULTS

In order to view the results, the figures are presented on a
weekly basis. The blue line indicates the experimental
measurements, whilst the green lines represent the
predictions made by the fuzzy system obtain for each
clustering techniques used.

Figure 1. Set of Temperature and Humidity prediction
using clustering and fuzzy techniques.


Figure 1a. Humidity (gr/m
3
) prediction using k-means clustering

Figure 1b. Temperature (C) prediction using k-means clustering.

105

Figure 1c. Humidity (gr/m
3
) prediction using c-means clustering

Figure 1d. Temperature (C) prediction using c-means clustering



Figure 1e. Humidity (gr/m
3
) prediction using subtractive clustering

Figure 1f. Temperature (C) prediction using subtractive clustering

VI. CONCLUSIONS AND FUTURE WORK

Fuzzy clustering techniques allow the automatic generation
of fuzzy models and can be utilized to modeling and predict
the behavior of greenhouse key parameters K-means, fuzzy
C-means clustering and fuzzy clustering subtractive are
useful techniques to describe complex dynamic systems and
provide an automatic way to generate robust fuzzy systems.
Future work will be involved with minimizing the system
error and try novel clustering techniques. Also, real-time
prediction may be carried out using sensor networks and
embedded systems.
REFERENCES

[1] J. Rodriguez Garza, E. Gorrostieta, R. Castaeda & G. Herrera,
Model for an Hydroponic Greenhouse, 2006 Proceedings of 12th
IEEE International Conference on Methods and Models in
Automation and Robotics MMAR.
[2] Gorrostieta, E., J.C. Pedraza, R.J. Carlos, Fuzzy Modelling of
Systems, Proceedings of 11th IEEE International Conference on
Methods and Models in Automation and Robotics MMAR, 29
August- 1 September 2005, Miedzyzdroje, Poland. ISBN 83- 60140-
85-5. (2005).
[3] Gorrostieta, E., Sotomayor, A., Aceves, M. A., Pedraza, C., and
Ramos, J. M., Modeling dynamic systems using ANFIS., 2009
Proceedings of the international Conference on Computational and
information Science, pp. 287-29.
[4] Paulo Salgado , J. Boaventura Cunha,Greenhouse climate
hierarchical fuzzy modelling 2005, Control Engineering Practice,
Vol 13, pp. 613-628.
[5] Takagi, T., and M. Sugeno, Fuzzy Identification of Systems and its
Application to Modeling and Control. 1985 IEEE Trans. Systems
Man and Cybernetics.15: 116- 132.
[6] Sugeno, M., and G. T. Kang, Structure Identification of Fuzzy
Model. 1998. Fuzzy Sets and Systems. 28: 15-33.
[7] T. Cox, E., The Fuzzy Systems Handbook: A Practitioners Guide to
Building, Using, and Maintaining Fuzzy Systems. Academic Press.
2002.
[8] Yager, R. and D. Filev, Generation of Fuzzy Rules by Mountain
Clustering, 1994, Journal of Intelligent & Fuzzy Systems, vol. 2,
No. 3, pp. 209- 219.
[9] Berkhin, P.: Survey of Clustering Data Mining Techniques. Grouping
Multidimensional Data, 2571, 2006.
[10] Cheng, Y.: Mean Shift, Mode Seeking, and Clustering. 1995, IEEE
Transactions on Pattern Analysis and Machine Intelligence 17(8),
790799.
[11] Comaniciu, D., Meer, P.: Mean Shift: A Robust Approach toward
Feature Space Analysis. 2002 IEEE Transactions on Pattern
Analysis and Machine Intelligence Vol. 24, No. 5, pp. 603619
[12] Tsoukalas, L., Uhrig, R.: Fuzzy and Neural Approaches in
Engineering. John Wiley & Sons, Inc., NY (1997).
[13] Yu, J.: General C-Means Clustering Model: A Unified Approach to
the Theories and Methods of Partitional Clustering Algorithms.
2003, Technical Report, AI Lab, Northern Jiaotong Univ., Beijing





106